Speaker
Description
The Jiangmen Underground Neutrino Observatory (JUNO) hosts the world's largest liquid scintillator detector, dedicated to measuring the neutrino mass ordering (NMO). The determination of NMO is highly sensitive to energy resolution, which is critically degraded by radioactive backgrounds, particularly from photomultiplier tubes (PMTs). JUNO employs 17,596 20-inch and 25,587 3-inch PMTs, whose glass components emit internal radioactivity generating non-negligible photoelectrons (PEs). These emissions contribute to background noise, distorting the reconstructed energy spectrum and impairing NMO sensitivity.
To address this challenge, we developed a novel photon-by-photon event reconstruction method based on a spatio-temporal Poisson model of PE density. By integrating a Markov Chain Monte Carlo (MCMC) framework with an Expectation Maximization (EM) algorithm, we probabilistically attribute each PE to one of three origins: vertex deposition, dark noise, or PMT radioactive emission. During MCMC iterations, the algorithm calculates the posterior probability of PE sources, enabling precise suppression of PMT-induced noise.
The proposed technique is anticipated to be applied to JUNO raw data, thereby enhancing the accuracy of NMO determination. Its framework also offers a generalizable approach for high-precision experiments reliant on large-scale PMT arrays.